1. Field of Invention
The invention relates generally to surface acoustic wave transducers, and more particularly to transducers using string weighting for impedance transformation with improved rejection properties.
2. Background Art
The use of surface-acoustic-wave (SAW) devices as filters or resonators is well known for having the advantages of high Q, low series resistance, small size, and good frequency-temperature stability when compared with other frequency control methods such as LC circuits, coaxial delay lines, or metal cavity resonators. As described in U.S. Pat. No. 4,600,905 to Fredricksen, typically, a SAW device contains a substrate of piezoelectric material such as quartz, lithium niobate, or zinc oxide. Input and output transducers are formed upon the substrate. The transducers convert input electrical signals to surface acoustic waves propagating upon the surface of the substrate and then reconvert the acoustic energy to an electric output signal. The input and output transducers are configured as interdigital electrode fingers that extend from pairs of transducer pads, with the electrodes alternately connected to the hot and the ground bus bars, which forms an unweighted transducer. Interdigital transducers may be formed by depositing and patterning a thin film of electrically conductive material upon the piezoelectric substrate.
Alternating electrical potential coupled to the input interdigital transducer induces mechanical stresses in the substrate. The resulting strains propagate away from the input transducer along the surface of the substrate in the form of surface acoustic waves. These propagating surface waves arrive at the output interdigital transducer, where they are converted to electrical signals.
The basic tapered transducer has been reported in the literature and in particular in an article by P. M. Naraine and C. K. Campbell (xe2x80x9cWideband Linear Phase SAW Filters Using Apodized Slanted Finger Transducers,xe2x80x9d Proceedings of IEEE Ultrasonics Symposium, Oct. ""83, pp 113-116). Naraine et al. discuss wide-band linear-phase SAW filters using apodized slanted or tapered finger transducers. In earlier publications, tapered finger transducer geometries have all the transducer fingers positioned along lines that emanate from a single focal point. A performance improvement was shown in U.S. Pat. Nos. 4,635,008 and 4,08,542 to Solie, the inventor of the present invention, by using hyperbolically tapered electrodes.
In Solie ""008 a dispersive SAW filter comprises hyperbolically tapered input and output transducers that are aligned such that normals from the transducers to a dispersive reflective array are aligned at substantially the same angle. The dispersive reflective array includes a multiplicity of parallel conductive strips or grooves formed in the device substrate on which the transducer rests. Constant spacing between the transducer fingers causes a relatively narrow band of frequencies to be generated by the input transducer and received by the output transducer. In Solie ""542, a reduction in the resistive loss associated with the long narrow electrodes in wide acoustic aperture devices is sought; a hyperbolically tapered transducer is provided with fingers having configuration paths that are subdivided into patterns that segment the acoustic beam width. Further disclosed is a means of transforming the impedance and thus reducing the insertion loss by a division of the SAW transducer structure into a plurality of subtransducers.
The use of tapered finger geometries on both input and output transducers permits the transduction of a wide range of surface acoustic wavelengths from input to output transducer, and thus provides an electrical filter with a wide frequency passband. Typically, high-frequency components are transduced in the regions of the transducer where the finger-to-finger distance is the least. Low-frequency components are transduced in the regions of the transducer where the finger-to-finger distance is the greatest. At any given frequency, a surface wave may be transmitted or received in a limited portion of the total acoustic aperture, and the width of this active portion is called the xe2x80x9ceffective aperturexe2x80x9d of the SAW beam.
The Naraine article states that for filters employing tapered finger transducer geometries, where the electrodes or fingers are straight-line segments emanating from a single point, there is an inherent negative slope of the amplitude response with increasing frequency, as large as 5 dB for a 50% bandwidth case reported in the IEEE article. Naraine""s article describes a method of flattening the amplitude-response curve of a tapered finger filter by utilizing finger apodization. Apodization is a technique in which the length of individual transducer fingers is selectively adjusted so that the overlap between fingers of opposite polarities changes along the path traveled by the surface acoustic wave.
A xe2x80x9ctapxe2x80x9d is defined as the gap between two adjacent electrodes. The strength of the tap, or the relative tap weight, is proportional to the voltage difference between the two electrodes. If each gap is surrounded by two electrodes of opposite polarity (hot and ground), then every tap has its maximum tap strength and has a relative tap weight of 1.0. This tap weighting is acceptable for coupling the strongest SAW in the shortest possible length of transducer; however, this configuration is not useful in making a filter with out-of-band rejection. In order to achieve useful rejection levels, the tap weights should assume predetermined values varying, for example, from 0.0 to 1.0. The technique used to achieve relative tap values is called tap weighting. Tapered transducers also need to be weighted for maximum rejection. One technique, as described above, apodization, is not applicable to tapered transducers. Three previously known techniques that are applicable to tapered transducers include withdrawal weighting, block weighting, and linewidth weighting. Linewidth weighting, wherein the electrode width is varied, cannot achieve an acceptable range in tap values owing to fabrication considerations, giving relative tap weights of approximately 0.8 to 1.0.
Another objective of the weighting technique is impedance transformation. Typically tapered transducers have very low impedance, resulting in a high insertion loss owing to the impedance mismatch between the source, which is typically 50 xcexa9, and the transducer, which may be  less than 1 xcexa9. The impedance values of tapered transducers are low because they have wide apertures to reduce diffraction, and the impedance is inversely proportional to the aperture width. Second, tapered transducers are typically long, several hundred wavelengths or even over one thousand wavelengths. Since all taps are connected in parallel across the two bus bars, this results in a very low impedance. Withdrawal weighting, which includes the elimination of some electrodes in the previously described alternating hot-ground connection pattern, does not increase the impedance.
The last of the previously known transducer configurations, block weighting, does provide a means for increasing impedance. An exemplary unweighted transducer 80 of length N (see FIG. 1) comprises a number, here 7, of subtransducers 81-87, all in the same acoustic path and phased so that all the subtransducers are acoustically in phase with each other. This depends upon the electrical connection as well as the spacing between subtransducers. The subtransducers are arranged in strings 88-90. The subtransducers within a string are connected in series, and the strings are connected in parallel across the two major bus bars 91,92. There is always an odd number of subtransducers in a string. The block-weighted transducer 80 of FIG. 1 has 1 or 3 subtransducers per string.
Each subtransducer 81-87 in FIG. 1 is represented by a capacitive impedance element 81xe2x80x2-87xe2x80x2 in FIG. 2, which shows how the voltage division within a string determines effective tap weights. The voltage across the transducer major bus bars 91,92 is now divided between the subtransducers depending upon the impedances of the three subtransducers 81-83. Since the impedance of each subtransducer is proportional to 1/ni and the sum of the voltages of the three subtransducers is 1.0, the tap weights within the subtransducers are wi=(1/ni)(1n1+1/n2+1/n3). Also, w1+w2+w3=1. Tap weighting is determined by choosing appropriate values of ni. Tap weights cannot be varied within a subtransducer, making the weighting technique inherently coarse, with the rejection provided less than ideal but still useful. This is illustrated in FIG. 3, wherein a target or desired weighting function 96 is shown by the continuous curve and the result achieved by block weighting is the stepped response 97.
The set of subtransducers connected in series, i.e., the string, has two ends that are connected to the two major bus bars, and it is possible to connect several strings to a single pair of major bus bars. Typically three or four strings are used in most tapered transducers. The tap weights are determined by varying the number of subtransducers in a string and then by varying the number of taps or wavelengths within each subtransducer. The impedance transformation can be illustrated by considering the limiting case where n1=n2=n3. Each subtransducer is only ⅓ as long as the original transducer with only ⅓ as many taps; so the impedance of each subtransducer is 3 times that of the original transducer. The string of three subtransducers has an impedance three times greater than one subtransducer; so the impedance of the string is 9 times greater than the original transducer. In general, if a single transducer is divided into a string of N equal subtransducers, then the impedance of the string is N2 greater than the original transducer. Since a block-weighted transducer uses strings of subtransducers, the use of block weighting increases the transducer impedance significantly, with the amount depending upon the design of the transducer and, in particular, how many subtransducers are used in a string.
It is an object of the present invention to provide a weighted surface-acoustic-wave transducer with improved rejection properties.
It is another object to provide such a transducer for providing impedance transformation.
These and other objects are provided by the present invention, a weighted surface-acoustic-wave transducer
String weighting is similar to block weighting; however, there are important differences. In string weighting there are a plurality of strings in a transducer, the plurality comprising the number of different tap weights within the transducer. In string weighting every subtransducer is preferably of minimal length, for example, nominally one wavelength, and there is an odd number of subtransducers in each string. The number must be odd, because the ends of the string must attach to the two bus bars. If there were an even number of subtransducers, the ends of the string would attach to the same bus bar, and there would be no voltage applied to the string. Since the full bus bar voltage is divided equally between the N subtransducers, the voltage at each tap is 1/N, and the length of the string is nominally N+1 wavelengths. Therefore, for a length of the transducer all the taps will have the same tap weight within the length of each string. This results in some degree of coarseness to the tap weighting because ideally the tap weights would be expected to change from tap to tap. However, the levels that they can assume provide at least several intermediate values between 0 and 1 with nothing in between, and this results in a degree of coarseness for withdrawal weighting, but here the taps can be changed from 0 to 1 at each tap. The net result is that the degree of coarseness for withdrawal weighting and for string weighting are about the same. Block weighting is, however, more coarse than string weighting, and as a result, string weighting can achieve better rejection.